- Joined
- Jun 21, 2002
No worries, we're all here to learn.
trents: Is there a specific benefit to having the voltage lower itself at idle? I guess in previous discussions of manual OC, I was under the impression that it was not a big deal because current still remained low at low loads.
I did some quick checks of behavior on my board, just for reference so I understand better what we're discussing. There were some reporting discrepancies between Vcore in the motherboard section of HWiNFO64 and "CPU Core Voltage" reported in the CPU section. The latter matched the oputput voltage reported from the VR. I ran a few minutes of small FFTs just to provide a load vs idle voltage comparison.
First I did a manual OC to 40x (4.0GHz) and left vcore and LLC on auto. Vcore reported 1.184V Max and 0.368V Min. CPU Core Voltage reported 1.181V Max and 1.144V Min.
Next was 40x with -0.05V offset, LLC remains auto. This caused a freeze, but I was still able to record data. Vcore reported 1.136V Max and 0.320V Min. CPU Core Voltage reported 1.131V Max and 1.119V Min.
Last I did 40x with vcore set to 1.2v, LLC auto. Vcore reported 1.2V Max and 1.184V Min. CPU Core Voltage reported 1.2V Max and 1.187V Min.
The undervolting reported in Vcore with Auto voltages is gone with a set voltage. However this undervolt was never reported in "CPU Core Voltage" which is what the VRMs spits out, so I'm not sure what is really going on with the undervolting. All of the minimum values for "CPU Core Voltage" reflect vdroop under load, rather than idle values. Meanwhile the minimum values for Vcore while voltage is set to auto reflect idle readings and only with a set voltage do they also reflect vdroop.
I did this test because I was not sure how functional automatic voltage would be with a set multiplier. Maybe next I should compare auto voltage and a set vcore at exactly VID at idle, and see if there is a difference in power draw form the wall. I did not think to measure that the first time.
